✨config ModelEngine Service

backend/consts/model.py

@@ -66,6 +66,7 @@ class ProviderModelRequest(BaseModel):
     provider: str
     model_type: str
     api_key: Optional[str] = ''
+    base_url: Optional[str] = ''
 
 
 class BatchCreateModelsRequest(BaseModel):

backend/services/model_health_service.py

@@ -195,14 +195,17 @@ async def verify_model_config_connectivity(model_config: dict):
             connectivity = await _perform_connectivity_check(
                 model_name, model_type, model_base_url, model_api_key, ssl_verify
             )
-
+            if not connectivity and ssl_verify:
+                connectivity = await _perform_connectivity_check(
+                    model_name, model_type, model_base_url, model_api_key, False
+                )
             if not connectivity:
                 return {
                     "connectivity": False,
                     "model_name": model_name,
                     "error": f"Failed to connect to model '{model_name}' at {model_base_url}. Please verify the URL, API key, and network connection."
                 }
-            
+
             return {
                 "connectivity": True,
                 "model_name": model_name,

backend/services/model_management_service.py

@@ -45,7 +45,9 @@ async def create_model_for_tenant(user_id: str, tenant_id: str, model_data: Dict
                 model_base_url.replace(LOCALHOST_NAME, DOCKER_INTERNAL_HOST)
                 .replace(LOCALHOST_IP, DOCKER_INTERNAL_HOST)
             )
-
+        model_data['ssl_verify'] = True
+        if "open/router" in model_base_url:
+            model_data['ssl_verify'] = False
         # Split model_name into repo and name
         model_repo, model_name = split_repo_name(
             model_data["model_name"]) if model_data.get("model_name") else ("", "")
@@ -286,7 +288,7 @@ async def delete_model_for_tenant(user_id: str, tenant_id: str, display_name: st
             raise LookupError(f"Model not found: {display_name}")
 
         deleted_types: List[str] = []
-        
+
         # Check if any of the models is multi_embedding (which means we have both types)
         has_multi_embedding = any(
             m.get("model_type") == "multi_embedding" for m in models
@@ -343,24 +345,24 @@ async def list_models_for_tenant(tenant_id: str):
     try:
         records = get_model_records(None, tenant_id)
         result: List[Dict[str, Any]] = []
-        
+
         # Type mapping for backwards compatibility (chat -> llm for frontend)
         type_map = {
             "chat": "llm",
         }
-        
+
         for record in records:
             record["model_name"] = add_repo_to_name(
                 model_repo=record["model_repo"],
                 model_name=record["model_name"],
             )
             record["connect_status"] = ModelConnectStatusEnum.get_value(
                 record.get("connect_status"))
-            
+
             # Map model_type if necessary (for ModelEngine compatibility)
             if record.get("model_type") in type_map:
                 record["model_type"] = type_map[record["model_type"]]
-            
+
             result.append(record)
 
         logging.debug("Successfully retrieved model list")

backend/services/model_provider_service.py

@@ -85,19 +85,22 @@ async def get_models(self, provider_config: Dict) -> List[Dict]:
             List of models with canonical fields
         """
         try:
-            if not MODEL_ENGINE_HOST or not MODEL_ENGINE_APIKEY:
-                logger.warning("ModelEngine environment variables not configured")
+            model_type: str = provider_config.get("model_type", "")
+            host = provider_config.get("base_url")
+            api_key = provider_config.get("api_key")
+
+            if not host or not api_key:
+                logger.warning("ModelEngine host or api key not configured")
                 return []
 
-            model_type: str = provider_config.get("model_type", "")
-            headers = {"Authorization": f"Bearer {MODEL_ENGINE_APIKEY}"}
+            headers = {"Authorization": f"Bearer {api_key}"}
 
             async with aiohttp.ClientSession(
                 timeout=aiohttp.ClientTimeout(total=30),
                 connector=aiohttp.TCPConnector(ssl=False)
             ) as session:
                 async with session.get(
-                    f"{MODEL_ENGINE_HOST}/open/router/v1/models",
+                    f"{host.rstrip('/')}/open/router/v1/models",
                     headers=headers
                 ) as response:
                     response.raise_for_status()
@@ -130,9 +133,8 @@ async def get_models(self, provider_config: Dict) -> List[Dict]:
                         "model_type": internal_type,
                         "model_tag": me_type,
                         "max_tokens": DEFAULT_LLM_MAX_TOKENS if internal_type in ("llm", "vlm") else 0,
-                        # ModelEngine models will get base_url and api_key from environment
-                        "base_url": MODEL_ENGINE_HOST,
-                        "api_key": MODEL_ENGINE_APIKEY,
+                        "base_url": host,
+                        "api_key": api_key,
                     })
 
             return filtered_models

frontend/app/[locale]/models/components/model/ModelAddDialog.tsx

@@ -4,10 +4,10 @@ import { useTranslation } from "react-i18next";
 import { Modal, Select, Input, Button, Switch, Tooltip, App } from "antd";
 import { InfoCircleFilled } from "@ant-design/icons";
 import {
-  LoaderCircle, 
-  ChevronRight, 
-  ChevronDown, 
-  Settings 
+  LoaderCircle,
+  ChevronRight,
+  ChevronDown,
+  Settings
 } from "lucide-react";
 
 import { useConfig } from "@/hooks/useConfig";
@@ -169,6 +169,7 @@ export const ModelAddDialog = ({
     // Whether to import multiple models at once
     isBatchImport: false,
     provider: "modelengine",
+    modelEngineUrl: "",
     vectorDimension: "1024",
     // Default chunk size range for embedding models
     chunkSizeRange: [
@@ -306,6 +307,14 @@ export const ModelAddDialog = ({
   // Check if the form is valid
   const isFormValid = () => {
     if (form.isBatchImport) {
+      // If provider is ModelEngine, require the ModelEngine URL as well.
+      if (form.provider === "modelengine") {
+        return (
+          form.provider.trim() !== "" &&
+          form.apiKey.trim() !== "" &&
+          ((form as any).modelEngineUrl || "").toString().trim() !== ""
+        );
+      }
       return form.provider.trim() !== "" && form.apiKey.trim() !== "";
     }
     if (form.type === MODEL_TYPES.EMBEDDING) {
@@ -602,6 +611,7 @@ export const ModelAddDialog = ({
         isMultimodal: false,
         isBatchImport: false,
         provider: "silicon",
+        modelEngineUrl: "",
         vectorDimension: "1024",
         chunkSizeRange: [
           DEFAULT_EXPECTED_CHUNK_SIZE,
@@ -675,6 +685,21 @@ export const ModelAddDialog = ({
               <Option value="modelengine">{t("model.provider.modelengine")}</Option>
               <Option value="silicon">{t("model.provider.silicon")}</Option>
             </Select>
+            {/* ModelEngine URL input (only when provider is ModelEngine) */}
+            {form.provider === "modelengine" && (
+              <div className="mt-3">
+                <label className="block mb-1 text-sm font-medium text-gray-700">
+                  ModelEngine URL
+                </label>
+                <Input
+                  placeholder={t("model.dialog.placeholder.modelEngineUrl")}
+                  value={(form as any).modelEngineUrl}
+                  onChange={(e) =>
+                    handleFormChange("modelEngineUrl", e.target.value)
+                  }
+                />
+              </div>
+            )}
           </div>
         )}
 

frontend/hooks/model/useSiliconModelList.ts

@@ -32,14 +32,18 @@ export const useSiliconModelList = ({
   const getModelList = async () => {
     setShowModelList(true)
     setLoadingModelList(true)
-    const modelType = form.type === "embedding" && form.isMultimodal ? 
-        "multi_embedding" as ModelType : 
+    const modelType = form.type === "embedding" && form.isMultimodal ?
+        "multi_embedding" as ModelType :
         form.type
     try {
       const result = await modelService.addProviderModel({
         provider: form.provider,
         type: modelType,
-        apiKey: form.apiKey.trim() === "" ? "sk-no-api-key" : form.apiKey
+        apiKey: form.apiKey.trim() === "" ? "sk-no-api-key" : form.apiKey,
+        baseUrl:
+          form.provider === "modelengine" && form.apiKey.trim() !== ""
+            ? (form as any).modelEngineUrl || ""
+            : undefined,
       })
       // Ensure each model has a default max_tokens value
       const modelsWithDefaults = result.map((model: any) => ({
@@ -68,20 +72,29 @@ export const useSiliconModelList = ({
   }
 
   const getProviderSelectedModalList = async () => {
-    const modelType = form.type === "embedding" && form.isMultimodal ? 
-        "multi_embedding" as ModelType : 
+    const modelType = form.type === "embedding" && form.isMultimodal ?
+        "multi_embedding" as ModelType :
         form.type
     const result = await modelService.getProviderSelectedModalList({
       provider: form.provider,
       type: modelType,
-      api_key: form.apiKey.trim() === "" ? "sk-no-api-key" : form.apiKey
+      api_key: form.apiKey.trim() === "" ? "sk-no-api-key" : form.apiKey,
+      baseUrl:
+        form.provider === "modelengine" && form.apiKey.trim() !== ""
+          ? (form as any).modelEngineUrl || ""
+          : undefined,
     })
     return result
   }
 
   // Auto-fetch model list when batch import is enabled and API key is provided
   useEffect(() => {
-    if (form.isBatchImport && form.apiKey.trim() !== "") {
+    const requiresUrl =
+      form.provider === "modelengine"
+        ? ((form as any).modelEngineUrl || "").toString().trim() !== ""
+        : true;
+
+    if (form.isBatchImport && form.apiKey.trim() !== "" && requiresUrl) {
       getModelList()
     }
     // eslint-disable-next-line react-hooks/exhaustive-deps

frontend/public/locales/en/common.json

@@ -600,6 +600,7 @@
   "model.dialog.placeholder.name": "Enter model name as in request body",
   "model.dialog.placeholder.displayName": "Enter display name for the model",
   "model.dialog.placeholder.url": "Enter model URL, e.g. https://api.openai.com/v1",
+  "model.dialog.placeholder.modelEngineUrl": "Enter ModelEngine host URL, e.g. https://120.253.225.102:50001",
   "model.dialog.placeholder.url.embedding": "Enter model URL, e.g. https://api.openai.com/v1/embeddings",
   "model.dialog.placeholder.apiKey": "Enter API Key",
   "model.dialog.placeholder.maxTokens": "Enter maximum tokens",

frontend/public/locales/zh/common.json

@@ -601,6 +601,7 @@
   "model.dialog.placeholder.name": "请输入请求体中的模型名称",
   "model.dialog.placeholder.displayName": "请输入模型的展示名称",
   "model.dialog.placeholder.url": "请输入模型URL, 例如: https://api.openai.com/v1",
+  "model.dialog.placeholder.modelEngineUrl": "请输入 ModelEngine 主机地址，例如：https://120.253.225.102:50001",
   "model.dialog.placeholder.url.embedding": "请输入模型URL, 例如: https://api.openai.com/v1/embeddings",
   "model.dialog.placeholder.apiKey": "请输入API Key",
   "model.dialog.placeholder.maxTokens": "请输入最大Token数",

frontend/services/modelService.ts

@@ -136,6 +136,7 @@ export const modelService = {
     provider: string;
     type: ModelType;
     apiKey: string;
+    baseUrl?: string;
   }): Promise<any[]> => {
     try {
       const response = await fetch(
@@ -147,6 +148,7 @@ export const modelService = {
             provider: model.provider,
             model_type: model.type,
             api_key: model.apiKey,
+            ...(model.baseUrl ? { base_url: model.baseUrl } : {}),
           }),
         }
       );
@@ -202,6 +204,7 @@ export const modelService = {
     provider: string;
     type: ModelType;
     api_key: string;
+    baseUrl?: string;
   }): Promise<any[]> => {
     try {
       const response = await fetch(
@@ -213,6 +216,7 @@ export const modelService = {
             provider: model.provider,
             model_type: model.type,
             api_key: model.api_key,
+            ...(model.baseUrl ? { base_url: model.baseUrl } : {}),
           }),
         }
       );

test/backend/services/test_model_health_service.py

@@ -1,5 +1,3 @@
-from consts.exceptions import TimeoutException
-import asyncio
 import os
 import sys
 from unittest import mock
@@ -792,5 +790,3 @@ async def test_embedding_dimension_check_wrapper_value_error():
         mock_logger.error.assert_called_once_with(
             "Error checking embedding dimension: Unsupported model type"
         )
-
-

test/backend/services/test_model_management_service.py

@@ -307,6 +307,7 @@ async def test_create_model_for_tenant_success_llm():
             "base_url": "http://localhost:8000",
             "model_type": "llm",
         }
+        model_data['ssl_verify'] = False
 
         await svc.create_model_for_tenant(user_id, tenant_id, model_data)
 
@@ -316,6 +317,32 @@ async def test_create_model_for_tenant_success_llm():
         assert mock_create.call_count == 1
 
 
+@pytest.mark.asyncio
+async def test_create_model_for_tenant_open_router_disables_ssl():
+    """When base_url contains 'open/router' ssl_verify should be set to False."""
+    svc = import_svc()
+
+    with mock.patch.object(svc, "get_model_by_display_name", return_value=None), \
+            mock.patch.object(svc, "create_model_record") as mock_create, \
+            mock.patch.object(svc, "split_repo_name", return_value=("modelengine", "m")):
+
+        user_id = "u1"
+        tenant_id = "t1"
+        model_data = {
+            "model_name": "modelengine/m",
+            "display_name": None,
+            "base_url": "https://api.example.com/open/router/v1",
+            "model_type": "llm",
+        }
+
+        await svc.create_model_for_tenant(user_id, tenant_id, model_data)
+
+        # Ensure a single record created and ssl_verify was disabled
+        assert mock_create.call_count == 1
+        create_args = mock_create.call_args[0][0]
+        assert create_args["ssl_verify"] is False
+
+
 @pytest.mark.asyncio
 async def test_create_model_for_tenant_conflict_raises():
     svc = import_svc()
@@ -459,7 +486,7 @@ async def test_create_model_for_tenant_multi_embedding_sets_default_chunk_batch(
         mock_dim.assert_awaited_once()
         # Should create two records: multi_embedding and its embedding variant
         assert mock_create.call_count == 2
-        
+
         # Verify chunk_batch was set to 10 for both records
         create_calls = mock_create.call_args_list
         # First call is for multi_embedding
@@ -519,7 +546,7 @@ async def test_batch_create_models_for_tenant_other_provider():
     if not hasattr(svc.ProviderEnum, 'MODELENGINE'):
         modelengine_item = _EnumItem("modelengine")
         svc.ProviderEnum.MODELENGINE = modelengine_item
-    
+
     with mock.patch.object(svc, "get_models_by_tenant_factory_type", return_value=[]), \
             mock.patch.object(svc, "delete_model_record"), \
             mock.patch.object(svc, "split_repo_name", return_value=("openai", "gpt-4")), \
@@ -529,7 +556,7 @@ async def test_batch_create_models_for_tenant_other_provider():
             mock.patch.object(svc, "create_model_record", return_value=True):
 
         await svc.batch_create_models_for_tenant("u1", "t1", batch_payload)
-        
+
         # Verify prepare_model_dict was called with empty model_url for non-Silicon/ModelEngine provider
         call_args = svc.prepare_model_dict.call_args
         assert call_args[1]["model_url"] == ""  # Should be empty for other providers
@@ -618,7 +645,7 @@ def get_by_display(display_name, tenant_id):
         update_calls = [call for call in mock_update.call_args_list if call[0][0] == "id1"]
         if update_calls:
             assert update_calls[0][0][1] == {"max_tokens": 8192}
-        
+
         # Should NOT update model2 (max_tokens same) or model3 (new max_tokens is None)
         # Verify model2 and model3 were not updated
         model2_calls = [call for call in mock_update.call_args_list if call[0][0] == "id2"]